Currently, when dsiplaying a GenomicRanges instance, there are often many more columns to be displayed than can fit on a screen at once.
The result is that when the instance is printed to the screen, single entries appear over multiple rows so that all columns can be displayed, e.g.:
print(gr)
GRanges object with 6 ranges and 23 metadata columns:
seqnames ranges strand | source type score phase
<Rle> <IRanges> <Rle> | <factor> <factor> <numeric> <integer>
[1] LmjF.01 [ 3704, 4702] - | TriTrypDB gene <NA> <NA>
[2] LmjF.01 [ 5790, 7439] - | TriTrypDB gene <NA> <NA>
[3] LmjF.01 [ 9061, 11067] - | TriTrypDB gene <NA> <NA>
[4] LmjF.01 [12073, 12642] - | TriTrypDB gene <NA> <NA>
[5] LmjF.01 [15025, 17022] - | TriTrypDB gene <NA> <NA>
[6] LmjF.01 [18137, 18886] - | TriTrypDB gene <NA> <NA>
ID Name description size
<character> <character> <character> <character>
[1] LmjF.01.0010 LmjF.01.0010 hypothetical+protein,+unknown+function 999
[2] LmjF.01.0020 LmjF.01.0020 hypothetical+protein,+conserved 1650
[3] LmjF.01.0030 LmjF.01.0030 Kinesin-13+1,+putative+(KIN13-1) 2007
[4] LmjF.01.0040 LmjF.01.0040 hypothetical+protein,+unknown+function 570
[5] LmjF.01.0050 LmjF.01.0050 carboxylase,+putative 1998
[6] LmjF.01.0060 LmjF.01.0060 hypothetical+protein,+conserved 750
web_id molecule_type organism_name translation_table topology
<character> <character> <character> <character> <character>
[1] LmjF.01.0010 <NA> <NA> <NA> <NA>
[2] LmjF.01.0020 <NA> <NA> <NA> <NA>
[3] LmjF.01.0030 <NA> <NA> <NA> <NA>
[4] LmjF.01.0040 <NA> <NA> <NA> <NA>
[5] LmjF.01.0050 <NA> <NA> <NA> <NA>
[6] LmjF.01.0060 <NA> <NA> <NA> <NA>
localization Dbxref locus_tag Alias
<character> <CharacterList> <character> <CharacterList>
[1] <NA> LmjF.01.0010 321438052,389592307,LmjF1.0010,...
[2] <NA> LmjF.01.0020 321438053,389592309,LmjF1.0020,...
[3] <NA> LmjF.01.0030 KIN13-1,Kif-13-1,321438054,...
[4] <NA> LmjF.01.0040 321438055,389592313,LmjF1.0040,...
[5] <NA> LmjF.01.0050 321438056,389592315,LmjF1.0050,...
[6] <NA> LmjF.01.0060 321438057,389592317,LmjF1.0060,...
Parent Ontology_term Frame Comment identical
<CharacterList> <CharacterList> <character> <CharacterList> <character>
[1] <NA> <NA>
[2] <NA> <NA>
[3] <NA> <NA>
[4] <NA> <NA>
[5] <NA> <NA>
[6] <NA> <NA>
overlap
<character>
[1] <NA>
[2] <NA>
[3] <NA>
[4] <NA>
[5] <NA>
[6] <NA>
-------
seqinfo: 36 sequences from an unspecified genome; no seqlengths
A much nicer approach might be to instead adopt a dplyr-like adaptive approach which modifies the number of columns displayed by default such that they all fit nicely on the screen:
> tbl_df(as.data.frame(gff1))
Source: local data frame [6 x 28]
seqnames start end width strand source type score phase
(fctr) (int) (int) (int) (fctr) (fctr) (fctr) (dbl) (int)
1 LmjF.01 3704 4702 999 - TriTrypDB gene NA NA
2 LmjF.01 5790 7439 1650 - TriTrypDB gene NA NA
3 LmjF.01 9061 11067 2007 - TriTrypDB gene NA NA
4 LmjF.01 12073 12642 570 - TriTrypDB gene NA NA
5 LmjF.01 15025 17022 1998 - TriTrypDB gene NA NA
6 LmjF.01 18137 18886 750 - TriTrypDB gene NA NA
Variables not shown: ID (chr), Name (chr), description (chr), size
(chr), web_id (chr), molecule_type (chr), organism_name (chr),
translation_table (chr), topology (chr), localization (chr),
Dbxref (chr), locus_tag (chr), Alias (chr), Parent (chr),
Ontology_term (chr), Frame (chr), Comment (chr), identical (chr),
overlap (chr)
Above is what the output would look like on a fairly small console (~800px wide). For a larger screen, more of the columns would be displayed in a similar manner.
Although I am sure people will have different preferences, I personally find this much cleaner and more pleasant to work with.
Is there any chance of changing the default behavior of GenomicRanges in the future to behave more like this?
While I'm not enlightened enough to be a dplyr user, it seems at least as desirable to show the last columns, since in a typical workflow, one is adding columns and wants to review them easily. Perhaps we could extend the row-wise head...tail approach to the columns. Personally, I've never really been bothered by showing all the columns, so we should support configuration like we do for the row-wise strategy.